Node definitions are specified using tables (see Table 2).

Attributes are defined by providing the Attribute name and a value, or a description of the value.

References are defined by providing the ReferenceType name, the BrowseName of the TargetNode and its NodeClass.

• If the TargetNode is a component of the Node being defined in the table, the Attributes of the composed Node are defined in the same row of the table.
• The DataType is only specified for Variables; "[<number>]" indicates a single-dimensional array, for multi-dimensional arrays the expression is repeated for each dimension (e.g. [2][3] for a two-dimensional array). For all arrays the ArrayDimensions is set as identified by <number> values. If no <number> is set, the corresponding dimension is set to 0, indicating an unknown size. If no number is provided at all the ArrayDimensions can be omitted. If no brackets are provided, it identifies a scalar DataType and the ValueRank is set to the corresponding value (see OPC 10000-3). In addition, ArrayDimensions is set to null or is omitted. If it can be Any or ScalarOrOneDimension, the value is put into "{<value>}", so either "{Any}" or "{ScalarOrOneDimension}" and the ValueRank is set to the corresponding value (see OPC 10000-3) and the ArrayDimensions is set to null or is omitted. Examples are given in Table 1.

Table 1 – Examples of DataTypes

Notation	DataType	ValueRank	ArrayDimensions	Description
Int32	Int32	−1	omitted or null	A scalar Int32.
Int32[]	Int32	1	omitted or {0}	Single-dimensional array of Int32 with an unknown size.
Int32[][]	Int32	2	omitted or {0,0}	Two-dimensional array of Int32 with unknown sizes for both dimensions.
Int32[3][]	Int32	2	{3,0}	Two-dimensional array of Int32 with a size of 3 for the first dimension and an unknown size for the second dimension.
Int32[5][3]	Int32	2	{5,3}	Two-dimensional array of Int32 with a size of 5 for the first dimension and a size of 3 for the second dimension.
Int32{Any}	Int32	−2	omitted or null	An Int32 where it is unknown if it is scalar or array with any number of dimensions.
Int32{ScalarOrOneDimension}	Int32	−3	omitted or null	An Int32 where it is either a single-dimensional array or a scalar.

• The TypeDefinition is specified for Objects and Variables.
• The TypeDefinition column specifies a symbolic name for a NodeId, i.e., the specified Node points with a HasTypeDefinition Reference to the corresponding Node.
• The ModellingRule of the referenced component is provided by specifying the symbolic name of the rule in the ModellingRule column. In the AddressSpace, the Node shall use a HasModellingRule Reference to point to the corresponding ModellingRule Object.

If the NodeId of a DataType is provided, the symbolic name of the Node representing the DataType shall be used.

Nodes of all other NodeClasses cannot be defined in the same table; therefore only the used ReferenceType, their NodeClass and their BrowseName are specified. A reference to another part of this document points to their definition.

Table 2 illustrates the table. If no components are provided, the DataType, TypeDefinition and ModellingRule columns may be omitted and only a Comment column is introduced to point to the Node definition.

Each Node defined in this specification has ConformanceUnits defined in OPC 10000-7 that require the Node to be in the AddressSpace. If a Server supports a ConformanceUnit, it shall expose the Nodes related to the ConformanceUnit in its AddressSpace. If two Nodes are exposed, all References between the Nodes defined in this specification shall be exposed as well.

The relations between Nodes and ConformanceUnits are defined at the end of the tables defining Nodes, one row per ConformanceUnit. The ConformanceUnit is reflected with a Category element in the UANodeSet file (see OPC 10000-6).

The Nodes defined in a table are not only the Node defined on top level, for example an ObjectType, but also the Nodes that are referenced, as long as they are not defined in other tables. For example, the ObjectType ServerType defines its InstanceDeclarations in the same table, so the InstanceDeclarations are also bound to the ConformanceUnits defined for the table. The table even indirectly defines additional InstanceDeclarations as components of the top-level InstanceDeclarations, that are not directly visible in the table. The TypeDefinitions and DataTypes used in the InstanceDeclarations, and the ReferenceTypes are defined in their individual tables and not in the table itself, therefore they are not bound to the ConformanceUnits of the table. As another example, the OrderedListType references the GeneralModelChangeEventType, which is defined in another table, and therefore not bound to the ConformanceUnits of the table. In case of ambiguity, the Category element of the UaNodeSet file provides the precise information.

Table 2 – TypeDefinition Table

Attribute	Value
Attribute name	Attribute value. If it is an optional Attribute that is not set "--" will be used.
References	NodeClass	BrowseName	DataType	TypeDefinition	ModellingRule
ReferenceType name	NodeClass of the TargetNode.	BrowseName of the target Node.	DataType of the referenced Node, only applicable for Variables.	TypeDefinition of the referenced Node, only applicable for Variables and Objects.	Referenced ModellingRule of the referenced Object.
NOTE Notes referencing footnotes of the table content.
Conformance Units
Name of conformance unit, one row per conformance unit

Components of Nodes can be complex that is containing components by themselves. The TypeDefinition, NodeClass, DataType and ModellingRule can be derived from the type definitions, and the symbolic name can be created as defined in 4.1. Therefore those containing components are not explicitly specified; they are implicitly specified by the type definitions.